Model launchesOpenAI launches always-on Dots autonomous agents amid security and privacy concernsOpenAI unveiled Dots, always-running autonomous agents built on the GPT-6 Astra model, at its San Francisco developer conference.3 min read
Model launchesOpenAI makes Ultrafast mode available for GPT-6 Astra and unveils the Pro 500 packageOpenAI announced that the Ultrafast performance mode is available today for the GPT-6 Astra model in Codex, ChatGPT Work, and via the API; support for GPT-6.1 Sol is coming soon.1 min read
Model launchesGPT-6.1 Sol has significantly improved alignment and safety compared to GPT-6 SolThe GPT-6.1 Sol model shows a substantial advance over GPT-6 Sol in alignment evaluations and has moved closer to the performance of GPT-6 Astra.1 min read
Model launchesOpenAI DevDay 2026: live coverage from Fort MasonThe report covers the OpenAI DevDay 2026 event held at Fort Mason in San Francisco.1 min read
Model launchesOpenAI withholds GPT-6.1 'Astra' release after failing internal safety checksOpenAI has paused the planned release of its GPT-6.1 Astra model after it failed to meet the company's internal safety and alignment standards.4 min read
Model launchesClaude Opus 5.5 Shows Strong Multimodal and Explainer-Video Capabilities Amid Broad Model WaveAnthropic’s Claude Opus 5.5, released the week of Sept 24–25, 2026, is drawing attention for high marks on vision and creative-code generation tasks, notably explainer videos and browser animations.7 min read
Model launchesSeveral major AI model updates within a week: debate over the meaning of "pacing the frontier"Opus 5.5, Grok 4.7, GPT-6 Sol, GPT-6 Astra and Sonnet 5.5 all launched within a week; this rapid succession of releases has sparked debate about how it fits with the "pacing the frontier" principle.1 min read
Model launchesAnthropic releases Claude Sonnet 5.5 — faster, cheaper, now the free-tier model on claude.aiAnthropic today introduced Claude Sonnet 5.5, which the company says runs over 30% faster and costs up to 30% less for most tasks while being priced the same as Sonnet 5.3 min read
Model launchesSonnet 5.5: faster and cheaper improvement compared to Sonnet 5The Sonnet 5.5 model was introduced and achieves better results than Sonnet 5 on numerous benchmarks, in some cases significantly.1 min read
Model launches2026 so far: coding agents, Claws and a year of rapid AI upheavalIn 2026 generative models and coding agents crossed a practical threshold: models that once made frequent mistakes became reliable enough for day-to-day software creation, spawning a new category of personal ‘‘Claw’’ agents and a surge in local model capability.7 min read
Model launchesOpus 5.5 appears virtually less limited compared to Fable 5.1The statement says the difference between Opus 5.5 and Fable 5.1 is explained by Opus’s cost-effectiveness and a 50% usage restriction applied to Fable.1 min read
Model launchesMeta’s Muse Gains Traction as OpenAI and Anthropic Roll Out New ModelsMeta’s consumer AI agent Muse is reportedly surpassing early ChatGPT engagement and is being prepared for integration into smart glasses and a small Tamagotchi‑style device.2 min read
